Better intellisense information

Cyrus Blather Better intellisense information | Insomnia Cure

Cyrus Blather Better intellisense information | Insomnia Cure — Sat, 13 Jun 2009 09:47:30 GMT

PingBack from http://insomniacuresite.info/story.php?id=4139

re: Better intellisense information

Cyrus Najmabadi — Mon, 02 Aug 2004 15:31:00 GMT

David: These behaviors are always available through the keyboard. Specifically this would fall under the command:

Edit.QuickInfo

which you can bind to any keystroke.

re: Better intellisense information

David A. Mellis — Mon, 02 Aug 2004 15:28:00 GMT

What about making this behaviour configurable via macros or something like the keyboard mapping dialog? It's better not to have too many options in the options dialog, but it would be nice to expose these commands to the user somewhere. For instance, I might want to call the "Show defintion" function with ctrl-d so I don't have to take my hands from the keyboard. Similarly, in debug mode, I could display variable values when hovering but variable definitions when ctrl-hovering.

re: Better intellisense information

Cyrus Najmabadi — Mon, 02 Aug 2004 05:30:00 GMT

Ahrha: Why would that be idea? :-)

re: Better intellisense information

Cyrus Najmabadi — Mon, 02 Aug 2004 05:29:00 GMT

Addy: I'm not sure if I agree. Making it configurable is not necessarily the right choice in all circumstances. If there is a "right" think to do, then pick that over a choice. It's also not right if you're forcing hte user to change the choice over and over depending on what they're doing.

re: Better intellisense information

arhra — Sat, 31 Jul 2004 11:16:00 GMT

I've no idea how hard it would be, but for local variables, how about making it show the last assignment to it, rather than the initilialization? (of course, ideally, the last assignment would _always_ be the initialisation... </functional zealot>

re: Better intellisense information

Addy Santo — Sat, 31 Jul 2004 05:54:00 GMT

This genre of questions have one universal answer: MAKE IT CONFIGURABLE. Now the question becomes what should the default behavior be, and I would say: disabled while the application is running (ie debug mode), otherwise enabled.

re: Better intellisense information

Cyrus Najmabadi — Sat, 31 Jul 2004 03:22:00 GMT

Brian: Great question. I wanted to address a few of your points:
A) VB does have a background compiler which does allow them to determine a lot of things as you type that we are (currently) unable to.
B) VB has worked a lot on scalability so that they can deal with large projects
C) This work I did does depend on any compiler features. It literally goes and grabs the initialization text where the item is defined and just jams it in :-)

In C# we have a separate compiler that we invoke when you hit "build" an we have something called "The C# Language Service" (LS) which you can think of as a trimmed down compiler. Where the compiler is built to be absolutely correct to the C# spec at any cost, the LS is defined to be as good as possible given the constraint of needing to be fast enough that we never interfere with typing. It's my complete belief that the two goals are not orthogonal. However, currently that's how it's architected and in the future we'll have to see about doing things different so that you can get the benefit's that are possible with a full background compiler.

It's important to realize that the LS is a mini-compiler in a sense. So any of the features that VB has could be done in C# if we're willing to extend this compiler to understand those ideas. Implementing them is usually no problem, the question is can we do it in a performant manner, can we test it adequately, and is the benefit high enough for the user versus the other features we can work on. I posted earlier about adding support for method and operator overload resolution. This is an area where VB already has functionality. Adding it would be possible, however it could drastically affect performance (now intellisense would be evaluating types of arguments to functions when binding them, which means abitrary expression evaluation). So if w want to do it we'd need to know that it was really worth the high cost it would have for us in terms of implementation time, time to get performance up, and time to test.

Does that help clear things up?

re: Better intellisense information

Brian — Fri, 30 Jul 2004 21:53:00 GMT

I might be off my rocker but....

I thought that VB.NET had a lot of cool Intellisense features at the cost of a background compiler. And the background compiler is what hindered the language from being used in HUGE projects.

Is there background compiling going on here?